Ferrera and Wilson [(1990) Vision Research, 30, 273-287] reported veridical perception of the direction of motion of Type I plaids, whose component gratings span the resultant direction, but marked misperception of the direction of motion of Type II plaids, whose component gratings both lie on one side of the resultant direction. Because they failed to find any effect of component direction (angular) separation on this misperception, Ferrera and Wilson concluded that the misperception was not due to perceptual repulsion of component directions. We report that component direction repulsion does occur, that plaid direction misperception is tuned to component separation, with larger repulsions for smaller angles. It is concluded that there is no fundamental difference in direction coding for Type I and Type II plaids, and that Ferrera and Wilson failed to find a direction separation effect because the range of separations they used was insufficiently broad to detect the slope of the angular function.